Musicians balance compensation rights with GenAI innovation | TechTarget

Getty Images/iStockphoto Musicians balance compensation rights with GenAI innovation While musicians have long used AI technology, GenAI is a new threat to their livelihoods. An effort has emerged to compensate artists whose work GenAI systems are trained on. Jenn Anderson-Miller is a leading voice of the new movement to protect musicians’ rights to be compensated for the work generative AI systems are trained on. Anderson-Miller, CEO and co-founder of music licensing and technology company Audiosocket, maintains that while AI is a useful tool, using copyrighted music to train GenAI systems without compensating musicians poses a threat to artists’ rights and livelihoods. This isn’t the first time musicians have had to fight to be compensated for their work. At the advent of the digital music era in the late 1990s, Napster enabled consumers to download songs from their favorite artists without paying. Major artists fought the platform, and a licensing system evolved. Today, streaming music platforms such as Spotify and Apple Music pay artists when users stream their music. Anderson-Miller is bullish on GenAI music applications, including Spotify’s new AI Playlist feature. Now in beta, it enables music consumers to compose custom playlists with text prompts. Audiosocket is working on a GenAI-powered music search system as well. But finding a fair balance between tech innovation and artists’ rights in the fast-evolving era of GenAI remains a challenge. A landmark lawsuit filed by The New York Times against GenAI vendors Microsoft and OpenAI will likely establish to what extent vendors will have to pay creators to use their work for AI model training. The vendors have put forth the fair use argument. Here, Anderson-Miller discusses the legal battle musicians face, how they might be compensated and the ways GenAI can be used to improve how music is consumed. Editor’s note: This interview was edited for clarity and conciseness. Let’s touch on the open letter from 200 musicians to GenAI vendors calling for fair compensation for artists’ work to safeguard their intellectual property rights. You were involved in that. What are those musicians’ main fears about GenAI? Jenn Anderson-Miller: I think it’s very well understood and known that their music has been scraped. The concern is, nobody gave permission. This is not fair use. If someone trains a model based on these famous artists’ [works], and then they go and monetize it all without the permission of the creator of the copyrighted works, that’s in violation of copyright law. That’s what is getting battled right now in the courthouses. And that’s what that letter was all about. You need to be ethical and responsible. If you’re making artists a part of the ongoing revenue stream and you’re doing really cool, innovative things, with permission, and we’re at the table to set the terms, we’re excited. But if you’re just scraping sites and doing things without our permission, where our voices are not heard, that’s scary, and I’m sure ultimately, in the courts, will be found to be illegal. Jenn Anderson-Miller CEO and co-founder, Audiosocket You’ve said licensing is the direction this issue is heading in. How is that going to look? Anderson-Miller: They’re going to have to bring the artists to the table. Artists have to be part of the conversation. Artists, from what I have been hearing, are holding out, and for good reason. I think that when you have a likeness or even something original that personifies you, you’re going to protect that. Artists’ catalogs should not go into these training models unless it is done with very clear, equitable models. It seems like we’re at an inflection point now, where individuals like you are laying the foundation for how musicians affected by GenAI are compensated in the future. Are you optimistic or pessimistic, and why? Anderson-Miller: I’m optimistic because I think the music industry learned a lot from the Napster era. Unlike when the digitization of music rolled around, I’m not seeing people dragging their heels. I think everybody is jumping in and engaging. And that’s a really good thing. I’m optimistic because I find so many people deeply engaged in the conversation. I think there’s enough awareness of the legal battle shaping up that people are being less brazen than they have been in the past, where they would just go out and do things and ask for forgiveness later. People are taking a more thoughtful approach than what I’ve seen in the past. On a skeptical note, do you see any signs that OpenAI, Google, Anthropic or Mistral are even entertaining the idea of licensing? The vendors essentially ignored the open letter. Anderson-Miller: Yeah, but I guess the thing that I wonder about is, what are they ignoring? Or is there a platform yet? Where are the artists coming together? Those of us that are having all these conversations are actually a little bit ahead of what’s going on, in terms of shaping the community around it. Part of what I’m trying to do is figure out how to put that platform together. Who are the parties I need to engage so that big tech and small tech are coming together with [intellectual property] owners to shape this? That’s what I’m not seeing yet. On Audiosocket, you have technology that you use to help consumers figure out what they want to listen to. How does your own involvement with AI technology inform your position regarding GenAI and musicians’ rights? Anderson-Miller: We started using AI almost four years ago. … We’ve seen the power that [AI] delivers for our users in terms of timesaving. I describe it to people who haven’t photo-edited before — when you look at stock photos, you’re looking at hundreds, if not thousands, of images in a minute. With music, you have to click and listen to every single option. And that is insanely time-consuming. We did some surveys of over 5,000 people using our site, and the greatest pain point was the amount of time it takes to search for and find music. So, what led us to the path of AI was [asking], ‘How can we do it better?’ At least as it relates to traditional AI, I think it’s leveling the playing field for artists. I think all artists are being given equal opportunity when you’re talking about AI versus a human-curated mechanism. [Artists are] talking about the ways that they’re using AI to help them write faster, to take them in many directions quicker. So, yes, our own technology is definitely what led me down the path of AI. GenAI coming out and making such a ripple led me down the exploratory journey around how authors can be part of this conversation where use is ethical by permission, and equitable — where they become part of an ongoing revenue stream. How does the new Spotify AI Playlist feature affect your artists? What are the benefits and what’s the downside? Anderson-Miller: There’s not enough data yet to say specifically, but I think we’re going to see artists rising up quicker. AI is leveling the playing field by basically allowing the user to be in the driver’s seat, whereas historically, from radio to Spotify, before AI, all of these things are primarily based on curating. And a lot of that has to do with money coming in and what’s trending — and way less to do with the user who is specifically accessing that platform in that moment with their own unique desires. AI is going to make it so that users are getting more music that they find they love because the AI is going to learn that specific person. And for independent artists, I think we’re going to see a leveling out and more rising up, just like with TikTok, where there’s been so much more music surfacing that nobody had heard before. You also have your own playlist feature on your site. How does that work? Anderson-Miller: We use the data from trend reports such as the Spotify top 100. We look at TikTok. We look at all these different sources that tell us what music is trending. And then we effectively look at what’s being used, and we curate playlists based on those pieces of information. We are in the process of building AI to replace playlisting, too, and we think that that’s even better than what we’re doing. We’re also integrating a text prompt. So, while it might be really hard for a user to pick a specific genre, or even a mood — because sometimes it feels like two or three moods, or two or three genres — they’ll be able to use words to describe it. Shaun Sutner is senior news director for TechTarget Editorial’s information management team, driving coverage of artificial intelligence, unified communications, analytics and data management technologies. He is a veteran journalist with more than 30 years of news experience.